git.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Mike Hommey <mh@glandium.org>
To: Jeff King <peff@peff.net>
Cc: Alexander Kuleshov <kuleshovmail@gmail.com>,
	Junio C Hamano <gitster@pobox.com>,
	git@vger.kernel.org
Subject: Re: [PATCH] Makefile: do not compile git with debugging symbols by default
Date: Fri, 23 Jan 2015 00:09:14 +0900	[thread overview]
Message-ID: <20150122150914.GA20135@glandium.org> (raw)
In-Reply-To: <20150122130036.GC19681@peff.net>

On Thu, Jan 22, 2015 at 08:00:36AM -0500, Jeff King wrote:
> On Thu, Jan 22, 2015 at 06:50:37PM +0600, Alexander Kuleshov wrote:
> 
> > Standard user has no need in debugging information. This patch adds
> > DEBUG=1 option to compile git with debugging symbols and compile without
> > it by default.
> 
> This explanation is missing why it is beneficial _not_ to have the
> debugging information.
> 
> I expect the answer is "it makes the executable smaller". And that is
> true, but it gets smaller still if you run "strip" on the result:
> 
>   $ make CFLAGS= >/dev/null 2>&1 && wc -c <git
>   2424248
> 
>   $ make CFLAGS=-g >/dev/null 2>&1 && wc -c <git
>   4500816
> 
>   $ strip git && wc -c <git
>   2109200
> 
> So I am not sure who this is helping. If you are size-conscious, you
> should use strip, in which case the "-g" flag does not matter (and we
> even have "make strip" to help you).
> 
> Is there some other reason to avoid the debugging information?

Maybe this comes from the misconception that debugging information
changes the generated code, which, in fact, it doesn't.

  $ make CFLAGS=-g LDFLAGS=-Wl,--build-id=none >/dev/null 2>&1 && wc -c <git
  4432768
  $ strip --strip-debug git && wc -c < git
  2391120
  $ cp git git_
  $ make -j4 CFLAGS= LDFLAGS=-Wl,--build-id=none >/dev/null 2>&1 && wc -c <git
  2400192
  $ strip --strip-debug git && wc -c < git
  2391120
  $ diff -s git git_
  Files git and git_ are identical

LDFLAGS=-Wl,--build-id=none just avoids creating a .note.gnu.build-id
section containing a uuid that varies between builds. The 9k difference
between unstripped vs stripped for the no-debug-info case comes from the
removal of the few symbols for source file names (all the symbols from
readelf -s git | grep ABS).

Mike

  reply	other threads:[~2015-01-22 15:09 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-01-22 12:50 [PATCH] Makefile: do not compile git with debugging symbols by default Alexander Kuleshov
2015-01-22 13:00 ` Jeff King
2015-01-22 15:09   ` Mike Hommey [this message]
2015-01-22 16:51   ` Alexander Kuleshov
2015-01-22 16:53     ` Alexander Kuleshov
2015-01-22 17:36     ` Matthieu Moy
2015-01-22 18:35       ` Jeff King
2015-01-22 22:55         ` Mike Hommey
2015-01-27  8:43           ` David Aguilar

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20150122150914.GA20135@glandium.org \
    --to=mh@glandium.org \
    --cc=git@vger.kernel.org \
    --cc=gitster@pobox.com \
    --cc=kuleshovmail@gmail.com \
    --cc=peff@peff.net \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).